...as in Sir Karl Popper (1902 - 1994). He was an Austrian philosopher of science who rejected the traditional idea that science advances by observation, deduction, and proof. Instead, Popper held that knowledge advances through a creative process of developing theories which are filtered out by Falsifiability: they were assumed to be correct until proven wrong. In the 1930's the logical positivists treated him as a kindred spirit. They held that a statement was true only if it was verifiable (a sort of reverse way of saying the same thing). The weakness of this approach is that Popper felt that knowledge advancement was an evolutionary process.
Well... sort of.
What he meant was that the theories that survive are the ones that best help us survive and prosper. When a theory is tested, we consider how it fits with our overall belief system and reject the theory that is most expendable - the one with the least inductive evidence supporting it. Sort of like cross-pollinating and irradiating your roses to evolve the Black Rose that your aesthetic has desired. This is like adhering to a religion that's convenient.
That's not really how real science works. Falsifiability is crucial, yes. However, convenience - to convolve a theory to optimize our existing belief system - no. That's not seeking Truth. It's taking the road most traveled.
Thomas Kuhn (1922-1996) argued forcefully that science advances only through a process of observing anomalies that don't fit with existing theories... these inevitably lead through a "paradigm shift" to a newer theory. Quine and Duhem (1861-1916) also retrospectively laid out a weakness of Popper's approach - they were certain that hypotheses can never be falsifiable in isolation. Any scientific theory is really an interdependent set of theories and assumptions, where any anomalous observation can falsify a number of different hypotheses. Falsifiability proceeding this way is closer to how modern science works.
There are also hypotheses that can never be falsifiable by definition: the existence of God, the existence of a Multiverse, etc.
For this reason, String Theory isn't science, but instead could be better described as mathematical philosophy. Perfecting a Black Rose - creating something beautiful and intellectually satisfying. While the theory strives to match observed physics, it isn't testable. For some, it's become a new religion (see an earlier blog entry about Atheists Believe).
This is the sort of thing that gives philosophy a bad name in the first place: it's just another example of the un-testable how-many-angels-can-dance-on-the-head-of-a-pin sort of stuff that preoccupied religious thinkers before the Enlightenment... before the idea that you could - and should - actually test things to prove them.